Is more information always better? Or are there some situations in which more information can make us worse off? Good (1967) argues that expected utility maximizers should always accept more information if the information is cost-free and relevant. But Good's argument presupposes that you are certain you will update by conditionalization. If we relax this assumption and allow agents to be uncertain about updating, these agents can be rationally required to reject free and relevant information. Since there are good reasons (...) to be uncertain about updating, rationality can require you to prefer ignorance. (shrink)

The framework of Solomonoff prediction assigns prior probability to hypotheses inversely proportional to their Kolmogorov complexity. There are two well-known problems. First, the Solomonoff prior is relative to a choice of Universal Turing machine. Second, the Solomonoff prior is not computable. However, there are responses to both problems. Different Solomonoff priors converge with more and more data. Further, there are computable approximations to the Solomonoff prior. I argue that there is a tension between these two responses. This is because computable (...) approximations to Solomonoff prediction do not always converge. (shrink)

In her 1993 book Freedom within Reason, Susan Wolf discusses what she identifies as an asymmetry between the good and the bad: to qualify as doing good in a praiseworthy way, it is not necessary that one should have the ability to do otherwise, but in order to qualify as doing something bad in a blameworthy way, it is necessary that one has the ability to do otherwise. In this chapter, I relate this asymmetry between the good and the bad (...) that Wolf discusses to a range of other asymmetries between the good and the bad that various philosophers – both historical figures and contemporary philosophers – have identified. I discuss different ways of thinking about the relations among these supposed asymmetries between the good and the bad, identify different ways in which one can relate to these ideas in one’s philosophical thinking, and lastly also argue that the long history of philosophical reflections on asymmetries between the good and the bad is interesting to contemporary discussions within applied ethics. In particular, I relate these ideas about asymmetries between the good and the bad to contemporary discussions about moral responsibility – and possible gaps in responsibility – within the ethics of artificial intelligence. (shrink)

How do we ascribe subjective probability? In decision theory, this question is often addressed by representation theorems, going back to Ramsey (1926), which tell us how to define or measure subjective probability by observable preferences. However, standard representation theorems make strong rationality assumptions, in particular expected utility maximization. How do we ascribe subjective probability to agents which do not satisfy these strong rationality assumptions? I present a representation theorem with weak rationality assumptions which can be used to define or measure (...) subjective probability for partly irrational agents. (shrink)

Many of our sources of knowledge only afford us knowledge that is inexact. When trying to see how tall something is, or to hear how far away something is, or to remember how long something lasted, we may come to know some facts about the approximate size, distance or duration of the thing in question but we don’t come to know exactly what its size, distance or duration is. In some such situations we also have some pointed knowledge of how (...) inexact our knowledge is. That is, we can knowledgeably pinpoint some exact claims that we do not know. We show that standard models of inexact knowledge leave little or no room for such pointed knowledge. We devise alternative models that are not afflicted by this shortcoming. (shrink)

Chancy modus ponens is the following inference scheme: ‘probably φ’, ‘if φ, then ψ’, therefore, ‘probably ψ’. I argue that Chancy modus ponens is invalid in general. I further argue that the invalidity of Chancy modus ponens sheds new light on the alleged counterexample to modus ponens presented by McGee. I close by observing that, although Chancy modus ponens is invalid in general, we can recover a restricted sense in which this scheme of inference is valid.

Could groups ever be an understanding subject (an epistemic agent ascribed with understanding) or should we keep our focus exclusively on the individuals that make up the group? The way this paper will shape an answer to this question is by starting from a case we are most willing to accept as group understanding, then mark out the crucial differences with an unconvincing case, and, ultimately, explain why these differences matter. In order to concoct the cases, however, we need to (...) elucidate what it means to be attributed with understanding and what makes up an epistemic agent. I shall argue that it is abilities, above all, that guide our attributions of understanding. While it is true that understanding must go beyond single acts, this is not going beyond as in going behind them (to private occurrences which are impossible to discern and don’t themselves contribute anything of value), but beyond as in considering what people could and would do: their discernable and valuable abilities. To conceptualize whom the abilities belong to, I will specify what it means to be an epistemic agent. I shall argue that it is being a successful target of the epistemic stance. The epistemic stance (heavily inspired by Dennett’s intentional stance) forms an instrumental abstraction by attributing epistemic properties (i.e., beliefs, epistemic aims, problem-solving tactics) to an entity to explain or predict its behavior. Macro-systematicity (a higher-level pattern which a theory can exploit) is what makes the epistemic stance’s abstraction an explanatory powerful one (regardless of how that macro-systematicity is realized) and emergence (the lack of a straightforward mapping-relation between the micro and the macro) is what makes its power unique to a particular level of explanation. To end, I suggest two kinds of mapping-relations and walk through the shift from reducibility to emergence. (shrink)

How might emerging and future technologies—sex robots, love drugs, anti-love drugs, or algorithms to track, quantify, and ‘gamify’ romantic relationships—change how we understand and value love? We canvass some of the main ethical worries posed by such technologies, while also considering whether there are reasons for “cautious optimism” about their implications for our lives. Along the way, we touch on some key ideas from the philosophies of love and technology.

PurposeIn this article, we aim to present and defend a contextual approach to mathematical explanation.MethodTo do this, we introduce an epistemic reading of mathematical explanation.ResultsThe epistemic reading not only clarifies the link between mathematical explanation and mathematical understanding, but also allows us to explicate some contextual factors governing explanation. We then show how several accounts of mathematical explanation can be read in this approach.ConclusionThe contextual approach defended here clears up the notion of explanation and pushes us towards a pluralist vision (...) on mathematical explanation. (shrink)

The absence of meaningfulness in life is meaninglessness. But what is the polar opposite of meaningfulness? In recent and ongoing work together with Stephen Campbell and Marcello di Paola respectively, I have explored what we dub ‘anti-meaning’: the negative counterpart of positive meaning in life. Here, I relate this idea of ‘anti-meaningful’ actions, activities, and projects to the topic of death, and in particular the deaths or suffering of those who will live after our own deaths. Connecting this idea of (...) anti-meaning and what happens after our own deaths to recent work by Samuel Scheffler on what he calls ‘the collective afterlife’ and his four reasons to care about future generations, I argue that if we today make choices or have lifestyles that later lead to unnecessarily early deaths and otherwise avoidable suffering of people who will live after we have died, this robs our current choices and lifestyles of some of their meaning, perhaps even making them the opposite of meaningful in the long run. (shrink)

In this paper, we engage in dialogue with Jonathan Pugh, Hannah Maslen, and Julian Savulescu about how to best interpret the potential impacts of deep brain stimulation on the self. We consider whether ordinary people’s convictions about the true self should be interpreted in essentialist or existentialist ways. Like Pugh et al., we argue that it is useful to understand the notion of the true self as having both essentialist and existentialist components. We also consider two ideas from existentialist philosophy (...) – Jean-Paul Sartre and Simone de Beauvoir’s ideas about “bad faith” and “ambiguity” – to argue that there can be value to patients in regarding themselves as having a certain amount of freedom to choose what aspects of themselves should be considered representative of their true selves. Lastly, we consider the case of an anorexia nervosa-patient who shifts between conflicting mind-sets. We argue that mind-sets in which it is easier for the patient and his or her family to share values can plausibly be considered to be more representative of the patient’s true self, if this promotes a well-functioning relationship between the patient and the family. However, we also argue that families are well-advised to give patients room to figure out what such shared values mean to them, since it can be alienating for patients if they feel that others try to impose values on them from the outside. (shrink)

I explain the New Riddle of Induction (Goodman 1946, 1955) in very brief words.

Ramsey (1926) sketches a proposal for measuring the subjective probabilities of an agent by their observable preferences, assuming that the agent is an expected utility maximizer. I show how to extend the spirit of Ramsey's method to a strictly wider class of agents: risk-weighted expected utility maximizers (Buchak 2013). In particular, I show how we can measure the risk attitudes of an agent by their observable preferences, assuming that the agent is a risk-weighted expected utility maximizer. Further, we can leverage (...) this method to measure the subjective probabilities of a risk-weighted expected utility maximizer. (shrink)

It has long been known that memory need not be a literal reproduction of the past but may be a constructive process. To say that memory is a constructive process is to say that the encoded content may differ from the retrieved content. At the same time, memory is bound by the authenticity constraint which states that the memory content must be true to the subject's original perception of reality. This paper addresses the question of how the constructive nature of (...) visual memory can be reconciled with the authenticity constraint. In what respect and to what extent may the content of a visual memory differ from the original perceptual state while still adequately reflecting the subject's original perception? (shrink)

It is a mistake to think that we cannot be morally responsible for forgetting because, as a matter of principle, forgetting is outside of our control. Sometimes we do have control over our forgetting. When forgetting is under our control there is no question that it is the proper object of praise and blame. But we can also be morally responsible for forgetting something when it is beyond our control that we forget that thing. The literature contains three accounts of (...) the blameworthiness of forgetting over which the agent has no control—the tracing account, the liberalized awareness condition, and attributionism. Even though these are competing accounts of the blameworthiness of harmful forgetting they are compatible with one another. In particular, it is possible to come up with a position that endorses the tracing account for certain kinds of harmful forgetting and attributionism for other kinds of harmful forgetting. (shrink)

Reuben Hersh confided to us that, about forty years ago, the late Paul Cohen predicted to him that at some unspecified point in the future, mathematicians would be replaced by computers. Rather than focus on computers replacing mathematicians, however, our aim is to consider the (im)possibility of human mathematicians being joined by “artificial mathematicians” in the proving practice—not just as a method of inquiry but as a fellow inquirer.

Advancements in artificial intelligence and (social) robotics raise pertinent questions as to how these technologies may help shape the society of the future. The main aim of the chapter is to consider the social and conceptual disruptions that might be associated with social robots, and humanoid social robots in particular. This chapter starts by comparing the concepts of robots and artificial intelligence and briefly explores the origins of these expressions. It then explains the definition of a social robot, as well (...) as the definition of humanoid robots. A key notion in this context is the idea of anthropomorphism: the human tendency to attribute human qualities, not only to our fellow human beings, but also to parts of nature and to technologies. This tendency to anthropomorphize technologies by responding to and interacting with them as if they have human qualities is one of the reasons why social robots (in particular social robots designed to look and behave like human beings) can be socially disruptive. As is explained in the chapter, while some ethics researchers believe that anthropomorphization is a mistake that can lead to various forms of deception, others — including both ethics researchers and social roboticists — believe it can be useful or fitting to treat robots in anthropomorphizing ways. The chapter explores that disagreement by, among other things, considering recent philosophical debates about whether social robots can be moral patients, that is, whether it can make sense to treat them with moral consideration. Where one stands on this issue will depend either on one’s views about whether social robots can have, imitate, or represent morally relevant properties, or on how people relate to social robots in their interactions with them. Lastly, the chapter urges that the ethics of social robots should explore intercultural perspectives, and highlights some recent research on Ubuntu ethics and social robots. (shrink)

In this paper I argue in favor of the compatibility of semantic externalism with privileged self-knowledge by showing that an argument for incompatibilism from switching scenarios fails. Given the inclusion theory of self-knowledge, the hypothesis according to which I am having twater thoughts while thinking that I have water thoughts simply isn't a (entertainable) possibility. When I am on Earth thinking earthian concepts, I cannot believe that I am thinking that twater is wet for I don't have the concept of (...) twater available; so this concept cannot figure in any of my mental states. Analogously, when I am on Twin Earth, I cannot mistakenly believe that I am entertaining water thoughts. No matter how often I am switched between Earth and Twin Earth, I will never erroneously think that I am having water thoughts while in fact I am having twater thoughts and vice versa. Privileged self-knowledge is therefore immune to skeptical arguments from switching and twinning scenarios. (shrink)

The conception of subjectivity developed by the German philosopher Hermann Schmitz (1927-) is especially suitable for cross-cultural investigations because its foundations lie in human experiences that are basic and universal. The paper has two aims. Firstly, to give an outline of Schmitz’s theory. Secondly, to show its usefulness (and its limits) by interpreting some Greek and Indian philosophers which, at the same time, represent certain main approaches to the problem of subjectivity.

In diesem Aufsatz stelle ich eine neue Interpretation der Reinhold’schen Sprachphilosophie vor. Mein Ziel ist es zu erklären, wie Reinhold der Meinung sein konnte, seine Sprachphilosophie stelle, ebenso wie seine Elementarphilosophie, den Versuch dar, Kants Kritische Philosophie zu fundieren. Außerdem möchte ich zeigen, worin die philosophische Bedeutung von Reinholds Ansatz gegenüber den Sprachphilosophien seiner Zeitgenossen besteht.

The growth of self-tracking and personal surveillance has given rise to the Quantified Self movement. Members of this movement seek to enhance their personal well-being, productivity, and self-actualization through the tracking and gamification of personal data. The technologies that make this possible can also track and gamify aspects of our interpersonal, romantic relationships. Several authors have begun to challenge the ethical and normative implications of this development. In this article, we build upon this work to provide a detailed ethical analysis (...) of the Quantified Relationship. We identify eight core objections to the QR and subject them to critical scrutiny. We argue that although critics raise legitimate concerns, there are ways in which tracking technologies can be used to support and facilitate good relationships. We thus adopt a stance of cautious openness toward this technology and advocate the development of a research agenda for the positive use of QR technologies. (shrink)

The overwhelming majority of those who theorize about implicit biases posit that these biases are caused by some sort of association. However, what exactly this claim amounts to is rarely specified. In this paper, I distinguish between different understandings of association, and I argue that the crucial senses of association for elucidating implicit bias are the cognitive structure and mental process senses. A hypothesis is subsequently derived: if associations really underpin implicit biases, then implicit biases should be modulated by counterconditioning (...) or extinction but should not be modulated by rational argumentation or logical interventions. This hypothesis is false; implicit biases are not predicated on any associative structures or associative processes but instead arise because of unconscious propositionally structured beliefs. I conclude by discussing how the case study of implicit bias illuminates problems with popular dual-process models of cognitive architecture. (shrink)

Hilary Putnam has famously argued that “we are brains in a vat” is necessarily false. The argument assumes content externalism (also known as semantic externalism and anti-individualism), that is, the view that the individuation conditions of mental content depend, in part, on external or relational properties of the subject’s environment. Recently content externalism has given rise to the hypothesis of the extended mind, whereby mental states are not only externally individuated but also externally located states. This chapter argues that when (...) content externalism is combined with the extended mind hypothesis it is robbed of its anti-skeptical power. Given the extended mind hypothesis, the supercomputer and the envatted brain can be regarded as aspects of the extended mind of the evil scientist. On this view, the thought contents of the coupled brain–computer–scientist system do not differ from those of a normal human. But without a difference in thought contents Putnam’s anti-skeptical argument crumbles. (shrink)

Wie der Titel des Aufsatzes bereits signalisiert, werde ich dafür argumentieren, dass das Gettier-Problem ein genuines Problem ist, keines, das sich lediglich einer falschen Fragestellung verdankt. Versuche, das Gettier-Problem aufzulösen statt zu lösen, sind zum Scheitern verurteilt. In den ersten beiden Abschnitten wird eine Typologisierung von Gettier-Fällen vorgenommen und zwischen zwei Lesarten des Gettier-Problems unterschieden. Im dritten Abschnitt werden einige Auflösungsversuche des Gettier-Problems der kritischen Prüfung unterzogen. Der vierte Abschnitt diskutiert die reliabilistische Antwort auf das Gettier-Problem. Es wird gezeigt, dass (...) sich mit Hilfe des Reliabilismus nicht alle Gettier-Fälle ausschließen lassen. Im fünften Abschnitt wird eine neue Lösung des Gettier-Problems entwickelt. (shrink)

The idea that there is a “Number Sense” (Dehaene, 1997) or “Core Knowledge” of number ensconced in a modular processing system (Carey, 2009) has gained popularity as the study of numerical cognition has matured. However, these claims are generally made with little, if any, detailed examination of which modular properties are instantiated in numerical processing. In this article, I aim to rectify this situation by detailing the modular properties on display in numerical cognitive processing. In the process, I review literature (...) from across the cognitive sciences and describe how the evidence reported in these works supports the hypothesis that numerical cognitive processing is modular. I outline the properties that would suffice for deeming a certain processing system a modular processing system. Subsequently, I use behavioral, neuropsychological, philosophical, and anthropological evidence to show that the number module is domain specific, informationally encapsulated, neurally localizable, subject to specific pathological breakdowns, mandatory, fast, and inaccessible at the person level; in other words, I use the evidence to demonstrate that some of our numerical capacity is housed in modular casing. (shrink)

Inquiry, Volume 57, Issue 1, Page 55-96, February 2014.

This paper explains and defends Reinhold’s epistemology of disagreement. The concept of agreement is of central importance for Reinhold’s philosophy. He attempts to settle the most basic disputes among post-Kantian philosophers by offering intermediate positions that reconcile the seemingly incompatible views. Moreover, Reinhold argues for epistemic objectivism, that is, the thesis that a group of philosophers sharing the same information and respecting each other’s opinion may not reasonably disagree. If the members of such a group search for truth then they (...) must converge toward consensus. Disagreement is irrational. (shrink)

Receiving information about threats to one’s health can contribute to anxiety and depression. In contemporary medical ethics there is considerable consensus that patient autonomy, or the patient’s right to know, in most cases outweighs these negative effects of information. Worry about the detrimental effects of information has, however, been voiced in relation to public health more generally. In particular, information about uncertain threats to public health, from—for example, chemicals—are said to entail social costs that have not been given due consideration. (...) This criticism implies a consequentialist argument for withholding such information from the public in their own best interest. In evaluating the argument for this kind of epistemic paternalism, the consequences of making information available must be compared to the consequences of withholding it. Consequences that should be considered include epistemic effects, psychological effects, effects on private decisions, and effects on political decisions. After giving due consideration to the possible uses of uncertain information and rebutting the claims that uncertainties imply small risks and that they are especially prone to entail misunderstandings and anxiety, it is concluded that there is a strong case against withholding of information about uncertain threats to public health. (shrink)

A common view in the philosophy of mind and philosophy of psychology is that there is an ideally correct way of categorizing the structures and operations of the mind, and that the goal of neuroscience and psychology is to find this correct categorizational scheme. Categories which cannot find a place within this correct framework ought to be eliminated from scientific practice. In this paper, I argue that this general idea runs counter to productive scientific practices. Such a view ignores the (...) plurality of aims and goals that neuroscientists and psychologists have in studying mental phenomena, and the necessity of employing distinct classificatory frameworks to achieve them. (shrink)

This paper argues for an overlooked dimension in the metaphysical microstructure of knowledge. The connection between knowledge and truth is even deeper than generally acknowledged. Knowledge, I argue, supervenes not only on a specific (namely modal) relation between the proposition p’s truth and an agent’s belief that p, but also on specific relations between the proposition’s truthmaker and the belief’s justification-maker. S knows that p only if the states of affairs referred to by S’s reasons for believing that p are (...) identical with, causally related to, or grounded in the states of affairs that make p true. (shrink)

This paper argues that, given the representational theory of mind, one cannot know a priori that one knows that p as opposed to being incapable of having any knowledge states; but one can know a priori that one knows that p as opposed to some other proposition q.

My dissertation is about Bayesian rationality for non-ideal agents. I show how to derive subjective probabilities from preferences using much weaker rationality assumptions than other standard representation theorems. I argue that non-ideal agents might be uncertain about how they will update on new information and consider two consequences of this uncertainty: such agents should sometimes reject free information and make choices which, taken together, yield sure loss. The upshot is that Bayesian rationality for non-ideal agents makes very different normative demands (...) than ideal Bayesian rationality. (shrink)

A Bayesian mind is, at its core, a rational mind. Bayesianism is thus well-suited to predict and explain mental processes that best exemplify our ability to be rational. However, evidence from belief acquisition and change appears to show that we do not acquire and update information in a Bayesian way. Instead, the principles of belief acquisition and updating seem grounded in maintaining a psychological immune system rather than in approximating a Bayesian processor.

After presenting evidence about categorization behavior, this paper argues for the following theses: 1) that there is a border between perception and cognition; 2) that the border is to be characterized by perception being modular (and cognition not being so); 3) that perception outputs conceptualized representations, so views that posit that the output of perception is solely non-conceptual are false; and 4) that perceptual content consists of basic-level categories and not richer contents.

Except for a patina of twenty-first century modernity, in the form of logic and language, philosophy is exactly the same now as it ever was; it has made no progress whatsoever. We philosophers wrestle with the exact same problems the Pre-Socratics wrestled with. Even more outrageous than this claim, though, is the blatant denial of its obvious truth by many practicing philosophers. The No-Progress view is explored and argued for here. Its denial is diagnosed as a form of anosognosia, a (...) mental condition where the affected person denies there is any problem. The theories of two eminent philosophers supporting the No-Progress view are also examined. The final section offers an explanation for philosophy 's inability to solve any philosophical problem, ever. The paper closes with some reflections on philosophy 's future. (shrink)

“What is the structure of thought?” is as central a question as any in cognitive science. A classic answer to this question has appealed to a Language of Thought (LoT). We point to emerging research from disparate branches of the field that supports the LoT hypothesis, but also uncovers diversity in LoTs across cognitive systems, stages of development, and species. Our letter formulates open research questions for cognitive science concerning the varieties of rules and representations that underwrite various LoT-based systems (...) and how these variations can help researchers taxonomize cognitive systems. (shrink)

Rapid advances in AI-based automation have led to a number of existential and economic concerns. In particular, as automating technologies develop enhanced competency they seem to threaten the values associated with meaningful work. In this article, we focus on one such value: the value of achievement. We argue that achievement is a key part of what makes work meaningful and that advances in AI and automation give rise to a number achievement gaps in the workplace. This could limit people’s ability (...) to participate in meaningful forms of work. Achievement gaps are interesting, in part, because they are the inverse of the (negative) responsibility gaps already widely discussed in the literature on AI ethics. Having described and explained the problem of achievement gaps, the article concludes by identifying four possible policy responses to the problem. (shrink)

Foundational theories of mental content seek to identify the conditions under which a mental representation expresses, in the mind of a particular thinker, a particular content. Normativists endorse the following general sort of foundational theory of mental content: A mental representation r expresses concept C for agent S just in case S ought to use r in conformity with some particular pattern of use associated with C. In response to Normativist theories of content, Kathrin Glüer-Pagin and Åsa Wikforss propose a (...) dilemma, alleging that Normativism either entails a vicious regress or falls prey to a charge of idleness. In this paper, I respond to this argument. I argue that Normativism can avoid the commitments that generate the regress and does not propose the sort of explanation required to charge that its explanation has been shown to be problematically idle. The regress-generating commitment to be avoided is, roughly, that tokened, contentful mental states are the product of rule-following. The explanatory task Normativists should disavow is that of explaining how it is that beliefs and other contentful mental states are produced. I argue that Normativism, properly understood as a theory of content, does not provide this kind of psychological explanation, and therefore does not entail that such explanations are to be given in terms of rule-following. If this is correct, Normativism is not the proper target of the dilemma offered by Glüer-Pagin and Wikforss. Understanding why one might construe Normativism in the way Glüer-Pagin and Wikforss must, and how, properly understood, it avoids their dilemma, can help us to appreciate the attractiveness of a genuinely normative theory of content and the importance of paying careful attention to the sort of normativity involved in norm-based theories of content. (shrink)

Replicating or exceeding human intelligence, not just in particular domains but in general, has always been a major goal of Artificial Intelligence (AI). We argue here that “human intelligence” is not only ill-defined, but often conflated with broader aspects of human psychology. Standard arguments for replicating it are morally unacceptable. We then suggest a reframing: that the proper goal of AI is not to replicate humans, but to complement them by creating diverse intelligences capable of collaborating with humans. This goal (...) renders issues of theory of mind, empathy, and caring, or community engagement, central to AI. It also challenges AI to better understand the circumstances in which human intelligence, including human moral intelligence, fails. (shrink)

Whole Brain Emulation has been championed as the most promising, well-defined route to achieving both human-level artificial intelligence and superintelligence. It has even been touted as a viable route to achieving immortality through brain uploading. WBE is not a fringe theory: the doctrine of Computationalism in philosophy of mind lends credence to the in-principle feasibility of the idea, and the standing of the Human Connectome Project makes it appear to be feasible in practice. Computationalism is a popular, independently plausible theory, (...) and Connectomics a well-funded empirical research program, so optimism about WBE is understandable. However, this optimism may be misplaced. This article argues that WBE is, at best, no more compelling than any of the other far-flung routes to achieving superintelligence. Similarly skeptical conclusions are found regarding immortality. The essay concludes with some positive considerations in favor of the Biological Theory of consciousness, as well as morals about the limits of Computationalism in both artificial intelligence and the philosophy of mind more generally. (shrink)

This essay attempts to clarify the nature and structure of aliefs. First I distinguish between a robust notion of aliefs and a deflated one. A robust notion of aliefs would introduce aliefs into our psychological ontology as a hitherto undiscovered kind, whereas a deflated notion of aliefs would identify aliefs as a set of pre-existing psychological states. I then propose the following dilemma: one the one hand, if aliefs have propositional content, then it is unclear exactly how aliefs differ from (...) psychological states we already countenance, in which case there is no robust notion of aliefs; on the other, if aliefs just contain associative content, then they cannot do the explanatory work set out for them, in which case there is no reason to posit aliefs at all. Thus, it appears that we have little reason to posit the novel category of robust aliefs. (shrink)